MAGNETIC RESONANCE IN CHEMlSTRY, VOL. 31, 2-6 (1993)

NMR Spectroscopy of Paramagnetic Complexes Part 39*-Natural Compounds

Abundance *H NMR of Paramagnetic Sandwich

Janet Blumel, Peter Hofmann and Frank H. Kohlert Anorganisch-chemisches Institut, Technische Universitat Miinchen, D-8046 Garching, Germany

The methylated metallocenes (MeCp),M with M = V, Cr, Mn, Co, Ni (1-5) were investigated by 'H and 'H NMR spectroscopy at natural abundance. The 2HNMR signals are narrower by a factor of up to 30 compared with the corresponding 'H NMR signals, thus establishing a n inexpensive method and a general improvement of the NMR spectra of paramagnetic x-complexes. This has allowed the resolution of the signal splitting of Cp deuterons of 1 and 5 which could not be observed earlier in the 'H NMR spectra. The origin of the small (1 and 5) and large (2-4) signal splittings is discussed and related to an extended Huckel calculation. The relative magnitude of the signal splitting is reproduced, but the signal assignment had to be deduced from 'C NMR results. Primary isotope shifts of up to 4.0 ppm were found for 1, 2, 4 and 5. The much higher values for 3 (up to 16.6 ppm) reflect the combined influenceof the intrinsic isotope shift and the isotope effect on the spin crossover. KEY WORDS

Paramagnetic NMR Natural abundance 'H NMR calculation

INTRODUCTION The use of ,H instead of 'H NMR spectroscopy has led to considerable progress in the characterization of paramagnetic organometallic compounds. In general, the signal half-width at a given temperature, W,, is much smaller for 'H NMR signals, and for the metallocenes (C5H,),M and (C,D,),M the theoretical factor of W,('H)/W,('H) = 42 has been attained.' This effect has been applied to detect deuterons which are located so close to the paramagnetic metal centre of a complex that the corresponding 'H NMR signal is broadened beyond detection.' Further, the spin crossover and the dimerization of manganocene could be clarified by using 'H NMR ~pectroscopy.~ The line narrowing effect is, of course, related to an improved spectral resolution. In this work we wanted to test this for the bis(methyl-~5-cyclopentadienyl)metal derivatives (MeCp),M because, for the series M = V, Cr, Mn, Co, Ni (1-5), a surprising metal-dependent signal splitting arises. For instance, less 'H NMR signals than expected from symmetry have been observed when M = V and Ni, and it was unclear whether this originated from resolution problems or whether the theory had to be m ~ d i f i e d Because .~ of the low receptivity of the deuteron and the large signal width encountered for paramagnetic molecules, deuterium-enriched samples had to be used in previous investigation^.'-^ Here we demonstrate that the method is not necessarily expensive, because deuterium at the * For Part 38, see Ref. 3.

t Author to whom correspondence should be addressed. 0749-1 581/93/01O002-05 $07.50 0 1993 by John Wiley & Sons, Ltd.

'H signal narrowing 'H isotope effect EHMO

natural abundance level may be sufficient. Further, we demonstrate that the qualitative molecular orbital concept used for the general understanding of the NMR results of paramagnetic x-complexes is also instrumental in accounting for the peculiarities outlined below.

RESULTS The liquid 1,l'-dimethylmetallocenes 1-5 gave good ' H NMR spectra within an acceptable accumulation time. A typical example is the spectrum of 5 in Fig. 1, which was obtained after 3.5 h. This shows that less concentrated samples and/or a shorter recording time are adequate. The signal half-width W,('H) is considerably smaller than W,('H) for all compounds (cf. Table 1). In the case of 5 this leads to signals of D-2-5 that are well separated (Fig. 1, inset A), whereas only one signal

* I

B

I

200

100

0 PPM

,

I

-100

-200

I

Figure 1. 46 MHz 'H N M R spectrum of a mixture of (MeCp),Ni and (MeCp).Fe (St, internal standard) at 338 K. The scale is given arbitrarily relative to (MeCp),Fe. Inset: signals of D-2-5at (A) 46 MHz and (6)H-2-5 at 300 MHz expanded by a factor of 9.

Received 10 February 1992 Accepted (revised) 15 September 1992

NATURAL ABUNDANCE 'H NMR OF PARAMAGNETIC SANDWICH COMPOUNDS

I

I

I

300

200

4

2

I

PPH

0

100

Figure2. (Top) 300 M H z ' H and (bottom) 46 MHz 'H NMR spectra of a mixture of (MeCp),V and (MeCp),Fe (St, internal standard) at 338 K. Scale as in Fig. 1 and only for the 'H N MR spectrum.

3

Mn and Co (2, 3 and 4) it is large, whereas for M = V and Ni (1 and 5 ) it is very small. For the understanding of these results an approach is desirable that is applicable for all (MeCp),M compounds, with slight modifications depending on the metal. Such an approach is M O theory combined with perturbation theory arguments, as illustrated in Fig. 3. Although calculated for (MeCp),Ni, the scheme in Fig. 3B is useful for all (MeCp),M compounds because, as shown below, only el-type orbitals need be considered. Starting from the Cp n-orbitals, substitution by a methyl group lifts the orbital degeneracy and places the symmetric el orbital (+el) above the antisymmetric orbital (a-el) (Fig. 3A), similar to the situation in toluene.6 The same applies for the e, orbitals. When two [MeCpl- Iigands are arranged parallel to each other at the distance found in metallocenes we obtain with further level splitting. Although here [MeCp]: [and for (MeCp),M in Fig. 3B] the symmetry is C,h, the levels have been labelled as a, el and e2 for simplicity in order to maintain the relationship to unperturbed D , , or D,, metallocenes. An EHMO calculation provides a guide for these level splittings: [MeCpl-, el 0.22 eV, e2 0.12 eV; [MeCpIt- with a ligand separation of 3.60 A, el 0.1710.23 eV, e2 0.10/0.14 eV. Note that the el orbitals of [MeCpli- which are appropriate for the interaction with metal 3d orbitals still have s-e, above a-e l . The diagram in Fig. 3B only shows those 7-r-orbitals of [MeCplq- which interact with the metal valence AOs (3d, 4s, 4p). The result of the interaction is the well known two-above-three level pattern of metallocenes. The bonding el orbitals are also given in Fig. 3B, whereas levels of t~ symmetry and n-group MOs of [MeCpli-, which only interact with metal 4p AOs, have been omitted for clarity. ~

appears for H-2-5 at 200 M H z . ~The signals of H-2-5 could be partly resolved by changing from 200 to 300 MHz, but the additional effect of ,H signal narrowing is clearly visible in the inset in Fig. 1. For the corresponding vanadocene 1 only one signal could be observed for H-2-5 even at 300 MHz, whereas D-2/5 and D-3/4 were separated by 1.9 ppm in the 'H NMR spectrum (Fig. 2). The paramagnetic signal shifts dpara(,H)and, for comparison, dpara(lH)were determined directly by referencing relative to the corresponding signals of internal (MeCp),Fe. These are shown together with other data in Table 1. Since both the 'H and 'H NMR spectra were obtained from the same sample at the same temperature (see Experimental), concentration-dependent intermolecular5 and temperature effects on the shifts could be excluded, so that the shift differences between analogous 'H and ,H NMR signals are primary isotope effects Ay(,I1H) = Para('H)- Bpara(,H).

Cobaltocene. We have shown previously7 that if only one

DISCUSSION Origin of the signal splitting

There is a striking difference in the signal splitting of the Cp protons within the series (MeCp),M: for M = Cr,

unpaired electron is present in the el* set, as is the case in (MeCp),Co, it will strongly prefer the lower of the two el* components. Let us suppose that as in Fig. 3B it is the one containing the a-e, contribution of the ligand and the d,, metal AO. We then follow qualitatively the squared carbon 2p, orbital coefficients, which

Table 1. *IIH NMR data. for paramagnetic 1,l'-dimethylmetallocenes (M = V, Cr, Mn, Co, Ni; 1-5) at 338 K "'H -215

Compound

Nucleus

1

'H 'H 'H ZH 'Hb 'H" 'H ZH 'H 'H

2 3 4 5

dD"'

301.1 300.3 327.7 325.8 4.7 4.7 -60.8 -60.8 -221.2 -221.4

''2H-3/4

W

60"'"

2750 100 2330 71 3960 320 290 9 460 25

301.1 298.4 280.4 278.8 80.5 73.8 -41.8 -41.7 -223.0 -223.3

o.8 1.9

0 o.2

W

2750 94 2040 66 81 00 410 220 81 0 27

C1I2H,

A Y

2'7 1'6

6'7 -0.1

0'3

par'

109.2 107.3 31.9 31.3 144.4 127.8 14.4 13.9 181.1 177.1

W

71 0 26 340 18 10850 540 56 6 270 17

*Y 1.9

16.'

0.5 4.0

'Paramagnetic shifts, Pa'"(+0.1 ppm), and isotope shifts, AEara, in ppm, negative sign for shifts to low frequency; signal half-width W i n Hz. *0.5 ppm owing to overlapping or broad signals. *1 .O ppm owing to overlapping or broad signals.

J. BLUMEL, P. HOFMANN AND F. H. KOHLER

4

in the spin delocalization. Therefore, we have to add the

E eV

-6

.

-8. -10.

-12

'

-14

'

a

[MeCp]

-

[MeCp]

t-

yk X

k

E eV

Fe2

- 6 . -8

. 46

s-ei

I

-10 .

-12

.

-14

.

-16

.

Ni 2+

(MeCp)*Ni

[Me CP]

;-

Figure 3. (A) M O diagram for the x-orbital interaction of two [MeCpl- ligands at the distance found in Cp,Ni. For labelling, see text. (B) M O diagram for the important metal-ligand x interactions in (MeCp),Ni.

in simple cases are a good measure of the spin density and thus of the hyperfine coupling constants A( 'H)* and A(13C).9 Since A('H/13C) is proportional to the corresponding NMR signal shifts, the signal of H-2/5 should be much more shifted than that of H-3/4. This is in accord with the 13C NMR result showing dPBra(C-l) < 6Para(C-3/4)< 6parB(C-2/5)4c.7c and confirms the level ordering given in Fig. 3B. Nickelocene. Since the splitting of the el* orbitals is

small, two unpaired electrons are present in (MeCp),Ni and both the a-el and s-el ligand orbitals are engaged

ci2(C2 p,) values of the two orbitals, which yields very similar spin densities at all ring carbon atoms. An approximate idea is provided by the EHMO results for [MeCpl-: C-1 0.351, C-3/4 0.359, C-2/5 0.389 (if only one el* orbital were engaged the splitting would be much larger: C-1, C-3/4, C-2/5 with 0.351/0.230/0.052 for a-el* and with 0/0.129/0.337 for s-el*). Qualitatively this reflects very well the fact that positions 3/4 and 2/5 can be distinguished only by the more powerful 'H NMR spectroscopy. The signal assignment for D-2/5 and D-3/4 is simple when based on the 13C NMR results, and when the spin distribution associated with either the a-el* or the +el* orbital (Fig. 3) dominates the signal shifts. For (MeCp),Ni we found = 1536 (C-l), 1510 (C-3/4) and 1356 (C-2/5)lo dpara(13C) (note that we have since changed the sign convention). It follows that the s-el* orbital dominates and that the signal of D-3/4 is more shifted than that of D-2/5. The dominance of the s-el* orbital seems to be reflected in the results of an EHMO calculation on (MeCp),Ni (Fig. 3B) where we find a metal contribution of 30% for a-e,* and of 28.4% for s-e,* at 0.10 eV higher energy. The reason for this is that the overlap (a-e,/d,,) is larger than